Polarization images can yield higher contrast than intensity images, providing the opportunity for dramatically improved object identification. Furthermore, incorporation of a polarimeter into a detection system allows for the potential to ascertain the Stokes parameter elements of a scene, thereby giving a complete identification of the polarization state of light reflected or emitted from objects in the scene. From such an analysis, the spatially varying two-dimensional state of polarization (SOP) can be determined.
SOP analysis is a useful technique for object characterization and distinction, particularly for differentiating man made versus natural objects. This is particularly valuable in the thermal infrared; if objects in a scene are emitting close to the background temperature of the environment (i.e., they are close to thermal equilibrium with their environment), then thermal detection typically yields ambiguous results. Addition of polarimetry data can often significantly enhance images of such objects as polarimetry can supply information that is unavailable by intensity imaging. For example, typical long-wavelength infrared (LWIR) intensity images provide little indication of the presence of a vehicle in the shadows of trees, while a polarization image makes the presence of an automobile obvious due to polarization associated with the smooth surfaces of the automobile.
Current techniques for imaging polarimetry include rotating retarder polarimeters. Through a series of sequential measurements, the complete spatial distribution of Stokes parameters in a scene can be determined. This method has several significant limitations. Rotating parts can lead to vibrational and mechanical problems. Images of dynamic scenes can also contain polarization artifacts as a result of combining a series of measurements. Other problems are related to oversampling and spatial synchronization.
Some of the problems with rotating retarder imaging polarimetry can be addressed with “snapshot” systems that do not require dynamic components, but instead take advantage of spatial carrier fringes and Fourier reconstruction techniques in order to provide a complete polarization analysis of a scene. Examples of such approaches are described in Oka and Saito, “Snapshot complete imaging polarimeter using Savart plates,” Proc. SPIE 6295:629508 (2008) and Oka and Kaneko, “Compact complete imaging polarimeter using birefringent wedge prisms,” Opt. Exp. 11:1510-1519 (2003), both of which are incorporated herein by reference. These approaches use birefringent materials to produce polarization dependent phase differences to produce snapshot images.
One example of such a snapshot system is based on a pair of Savart plates (SPs) introduced in a collimated space in an imaging system. An SP shears incident radiation using crystal birefringence to produce two laterally displaced, orthogonally polarized beams. By combining two orthogonal SPs, an incident optical flux is sheared to create four separate beams. After transmission by an analyzer, these beams are recombined with a lens, resulting in amplitude modulated interference fringes containing state of polarization (SOP) information on the image plane.
While such SP systems are impressive in their snapshot capabilities, they suffer from significant limitations. Due to the reliance on interference effects, the temporal coherence of imaging radiation presents a constraint in that the visibility of the interference fringes is inversely proportional to the spectral bandwidth. For instance, in the LWIR (8-12 μm wavelengths), a fringe visibility of 50% at a mean wavelength of 10 μm requires limiting optical bandwidth Δλ50%≈373 nm, which is a significant constraint with respect to the signal to noise ratio (SNR) of the acquired data. In addition, SP polarimeters require SPs which can be expensive due to the birefringent crystals required. In many wavelength regimes, especially the infrared, the required large crystals (clear apertures >25 mm with thicknesses >10 mm) are either unavailable or prohibitively expensive. Moreover, materials suitable for LWIR use such as CdSe or CdS have birefringences B=|ne−no| that are approximately 10 times less than those of materials suitable for use at visible wavelengths. As a result, thick crystals are needed.
These birefringent material limitations can be avoided through the implementation of a reflective interferometric scheme. Mujat et. al., “Interferometric imaging polarimeter,” JOSA A:21:2244-2249 (2004), which is incorporated herein by reference, discloses an interferometric imaging polarimeter based on a modified Sagnac interferometer. In this system, a polarizing beam splitter is used to transmit an input beam into an interferometer, and a phase difference between orthogonal polarizations produced by displacing one of the mirrors in the interferometer is used to create an interference pattern. Irradiance measurements and coherence matrix techniques are then employed to determine the state of polarization from a set of two temporally spaced images. These methods are subject to similar registration problems that plague rotating retarder polarimeters for dynamic scenes. In addition, in some cases, conventional polarimeters produce images that require additional processing for viewing.